Suction buckets are large shell structures that have become a prominent alternative to pile foundations for bottom-fixed and floating offshore wind turbines. They are embedded by applying negative pressure, which leads to a high risk of structural buckling during the installation. The prediction of the buckling strength of such large shells is subject to uncertainty, since it depends significantly on the initial geometric imperfections resulting from the fabrication process. The aim of this work is to understand and reduce uncertainties in the determination of the buckling pressure. Previous work on suction buckets revealed that the choice of a representative imperfection form and amplitude is very challenging and has not yet been solved in a generalized manner. In this work, a stochastic modeling approach is introduced, which considers more realistic imperfection patterns. This approach is compared to widely established imperfection forms such as buckling mode affine imperfections and analytically described weld depressions. The generated imperfection patterns are applied to geometrically and materially nonlinear finite element models and the buckling pressures are calculated. By quantifying the impact of different imperfection forms and amplitudes, uncertainties can be reduced, and design optimization and cost minimization are enabled.

The prediction of the buckling resistance of shells is subject to uncertainty since it depends significantly on the initial geometric imperfections caused by the fabrication process. However, the load case of vacuum or uniform external pressure has not yet been extensively studied. Stress design approaches implement a reduction factor alpha that accounts for the effect of imperfections. This factor is assumed to be a general lower bound but does not account for different geometric ratios. However, it is to be expected that different imperfection sensitivities prevail for different cylinder lengths or length to thickness ratios. To better understand this aspect, a comprehensive numerical study of buckling resistances of imperfect cylindrical shells is conducted, which covers a range of different length, radius and thickness ratios with focus on thicker and shorter cylinders. Within this study, three different geometric imperfection forms are investigated: linear bifurcation eigenmodes, and ovalisation and weld depressions, which represent typical imperfections of manufacturing processes. Further, a formulation for longitudinal weld depressions is introduced, since this direction of the imperfection is more deleterious than axisymmetric circumferential weld depressions. The results of this comprehensive study contribute to understanding the effect of different imperfection forms and amplitudes on different cylinder dimensions.

Recently, suction buckets have become a very prominent foundation for bottom fixed and floating offshore wind turbines. They are embedded with an installation force that stems from water evacuation inside the bucket. This internal negative pressure leads to a high risk of structural buckling. The buckling strength is significantly reduced by geometric imperfections. In previous work, equivalent geometric imperfection forms were introduced and the lower bound was evaluated. However, it has not yet been possible to identify a generally appropriate imperfection form. A probabilistic design approach based on realistic imperfections was not yet considered for suction buckets. Therefore, in this work, a stochastic modeling approach is introduced, which bases on measured data. The imperfection is decomposed to the half-wave cosine Fourier representation. Realizations of the imperfection pattern are generated by filtering white noise with the amplitude spectrum. They are then applied as out of plane deviations on a geometrically and materially nonlinear finite element model and evaluated. The resulting buckling pressure distribution can then be evaluated for different reliability levels. By considering more realistic imperfections and a plastic soil model, the buckling pressure increases by up to a factor of two compared to the conservative stress-based buckling approach.